org-weights — Show how heavy Org subtrees are
org-weights is an Emacs minor mode for Org format files. It displays the weights for visible Org headers directly on the header line. The weights of a header are the counts of sub-trees and paragraphs for the the sub-tree starting with that header. Paragraphs include items and other equivalent structures.
To install org-weights, just copy
org-weights.el somewhere Emacs may
find it. Optionally, assign some key binding to toggle the mode. For
one, I added these lines to my
(autoload 'org-weights-mode "org-weights" nil t) (define-key org-mode-map "\C-cow" 'org-weights-mode)
yet of course, one may choose any other key binding.
The mode gets activated or deactivated with
M-x org-weights-mode RET.
Whenever this minor mode gets activated, all headers which are visible
then display, towards the right side of the window, a small region
which hides the contents beneath if any. That region contains a
sequence of asterisks acting as a reminder for the current header
level, then either one or two counters. One of them shows the number
of included items or paragraphs below the sub-tree of the Org file
under the current header, the other gives the number of sub-headers if
any, in which case that second number is shown within parentheses and
with the letter
h as a reminder that this is a header count.
The numbers are not shown for a header line when the edition cursor (the point in Emacs parlance) happens to be on that line, so to not interfere with line editing. As soon as the cursor moves out of a header line, the monitoring resumes for that line.
Under org-weights mode, these numbers get updated dynamically while the Org file is being edited. If new headers are added, or otherwise revealed after having been hidden, org-weights do not start tracking them: one may re-activate org-weights minor mode again for them to be monitored as well, or merely wander the cursor over the new headers.
It happened that I chose Org to write short manuals for my customers. Even if this is theoretically immaterial, I find a manual more comfortable when chapters, and sections within chapters, do not too wildly vary in size. Too big a section often means that the section has to be broken up in smaller chunks, and it is often possible to merge smallish sections together, given of course that some logic could be created to make such a merge reasonable.
[2012-02-26 dim] After I spoke on the Org mailing list of my need to see the weights of sub-trees on header lines, Nicolas Goaziou was kind enough to share a function for computing the weights. I merely wrapped some more Lisp code around his skeleton. At the time, a command was displaying the weights, which were all vanishing as soon as any modifying action was made to the Org file. This encouraged me at reaching the equilibrium I wanted in my manuals.
[2013-02-17 dim] Nowadays, I have hundreds of Org files containing notes of all kinds, which I often feel like cleaning or re-organizing when visiting them. It always helps having an overall idea of the work needed to revise or otherwise handle sections of an Org file, by getting an idea of the weight of each header. Knowing the energy you currently have, or how much time is available, I may select if I’ll tackle a bigger node or a smaller node. That’s a simple thing, but it helps me at better investing the bits of time I have. So, this mode keeps being useful to me. I more recently had the need to frequently re-evaluate the weights after modifications, and got tired of repeating the displaying command after each modification. So I adapted the code so the weights get updated instead of disappearing. Bastien also suggested that I turn my code into a small library: here it is.
org-weights works relatively OK for me. Bugs surely remain. Here are those I noticed:
- In case of a demotion, the previous header needs updating. Currently, after some part of an Org file changes, the weights of its header is revised, and recursively up to the root. However, after a demotion, the header that was immediately holding the demoted header is not in its up chain anymore, and so, it is not revised. An easy work around is to move the cursor on that wrong header and out, the weights then get recomputed.
- Cut and pasting seems to move overlays as well? I got a few spurious overlays through such operations. I guess I would have to study and understand better how it all works.
- While this code is fast enough for my needs, it uses a bit of quadratic time while scanning for headers, so it might become slow if there were a big number of headers. This could be alleviated if weights computed deeper down a tree were re-used without recomputing them anew to get the weights at an upper level of the hierarchy.
- Overlays are sometimes copied in unexpected ways, like while capturing. Maybe should org-weights be automatically inhibited while capturing, short of solving the real problem? (Sébastien Vauban, [2013-02-27 mar])